Due to the difficult, if not impossible, task of in-vivo dose measurements for the patient, accurate dose calculation is imperative in radiotherapy, as it gives the only means to estimate the actual dose delivered to the patient. It has been recommended that, for photon therapy, dose be calculated to about plus/minus 3% accuracy. Such standard is difficult and not yet fully attained. The complex nature of photon dose deposition in a heterogeneous medium dictates that, at present, all clinical methods are inevitably approximate. While the advent of x-ray computed tomograph (CT) and affordable high-performance computers offer potentials for improvement, to date, none of the existing algorithms can be applied to the wide range of energies and geometries encountered in radiotherapy and still satisfy the recommended accuracy guideline. At our institute in Washington University, St. Louis, we have an on-going collaborative effort with the Institute For Biomedical Computing to develop the "delta-volume" algorithm for accurate dose calculation and implement it for practical use. Proper dose calculations require both photon- and electron-transport considerations and must be based on fundamental understanding of the problem. Results from our recently completed photon-transport phase of the study show that, for cobalt-60 beams, the "delta-volume" algorithm is capable of achieving better than plus/minus 3% accuracy. More important, the algorithm forms a logical basis for extension to higher photon energies, thus a framework for an unified method for clinical dose calculations. In this proposal, we specifically wish to: (1) extend the present algorithm to all photon energies commonly used in radiotherapy by incorporating appropriate electron-transport calculation onto the present photon-transport model; (2) perform experimental measurements in basic geometries such that boundary conditions can be established and limits of approximations evaluated; and (3) evaluate the efficacy of the final algorithm in clinical realistic geometries and extreme phantom arrangements so that the range of application can be established. Results from this work will complement the computer hardware development carried out by the Institute For Biomedical Computings.